This study evaluated the tolerance and potential pharmacokinetic interactions between clarithromycin (500 mg every 12 h) and rifabutin (300 mg daily) in clinically stable human immunodeficiency virus-infected volunteers with CD4 counts of <200 cells/mm3. Thirty-four subjects were randomized equally to either regimen A or regimen B. On days 1 to 14, subjects assigned to regimen A received clarithromycin and subjects assigned to regimen B received rifabutin, and then both groups received both drugs on days 15 to 42. Of the 14 regimen A and the 15 regimen B subjects who started combination therapy, 1 subject in each group prematurely discontinued therapy due to toxicity, but 19 of 29 subjects reported nausea, vomiting, and/or diarrhea. Pharmacokinetic analysis included data for 11 regimen A and 14 regimen B subjects. Steady-state pharmacokinetic parameters for single-agent therapy (day 14) and combination therapy (day 42) were compared. Regimen A resulted in a mean decrease of 44% (P = 0.003) in the clarithromycin area under the plasma concentration-time curve (AUC), while there was a mean increase of 57% (P = 0.004) in the AUC of the clarithromycin metabolite 14-OH-clarithromycin. Regimen B resulted in a mean increase of 99% (P = 0.001) in the rifabutin AUC and a mean increase of 375% (P < 0.001) in the AUC of the rifabutin metabolite 25-O-desacetyl-rifabutin. The usefulness of this combination for prophylaxis of Mycobacterium avium infections is limited by frequent gastrointestinal adverse events. Coadministration of clarithromycin and rifabutin results in significant bidirectional pharmacokinetic interactions. The resulting increase in rifabutin levels may explain the increased frequency of uveitis observed with concomitant use of these drugs.